Growing evidence for separate neural mechanisms for attention and consciousness

Abstract

Our conscious experience of the world seems to go in lockstep with our attentional focus: We tend to see, hear, taste, and feel what we attend to, and vice versa. This tight coupling between attention and consciousness has given rise to the idea that these two phenomena are indivisible. In the late 1950s, the honoree of this special issue, Charles Eriksen, was among a small group of early pioneers that sought to investigate whether a transient increase in overall level of attention (alertness) in response to a noxious stimulus can be decoupled from conscious perception using experimental techniques. Recent years saw a similar debate regarding whether attention and consciousness are two dissociable processes. Initial evidence that attention and consciousness are two separate processes primarily rested on behavioral data. However, the past couple of years witnessed an explosion of studies aimed at testing this conjecture using neuroscientific techniques. Here we provide an overview of these and related empirical studies on the distinction between the neuronal correlates of attention and consciousness, and detail how advancements in theory and technology can bring about a more detailed understanding of the two. We argue that the most promising approach will combine ever-evolving neurophysiological and interventionist tools with quantitative, empirically testable theories of consciousness that are grounded in a mathematically formalized understanding of phenomenology.

Keywords: Attention: Neural Mechanisms; Cognitive neuroscience; Consciousness; Electrophysiology.

Figure 1.

fMRI and EEG evidence for separate neuronal processes underlying attention and consciousness. a) fMRI study showing that, while attention increases neuronal response magnitudes independently of whether a stimulus reached conscious perception, the larger difference in neuronal activity is related to whether a subject was conscious or unconscious of that stimulus (note the difference in units on the y-axis). b) EEG response magnitudes to conscious and unconscious stimuli in the presence or absence of attention, showing a similar effect as a). c) EEG responses to conscious (magenta) vs. unconscious attentional (cyan) cues. Note that even a cue that is not consciously perceived evokes an elevated (albeit somewhat smaller). EEG response, indicating attentional enhancement of neuronal responses.

Figure 2.

Double dissociations between attention and consciousness demonstrated using E/MEG. a) EEG responses to tactile stimuli that were either consciously perceived or not (y-axis) and at the same time either attended or not (x-axis, red vs. blue). Note that each of these combinations yielded a different EEG response magnitude. b) Spectral profiles of MEG responses to consciously perceived vs. unconscious visual stimuli (left), independent of attentional state and attended vs. unattended stimuli, independent of conscious state (right). Note the difference in frequency and onset for the power maxima of each of the two contrasts.

Figure 3.

a) Schematic summary of the effects of attention and conscious state on the magnitude of large-scale sensory brain responses (modified from (Koivisto & Revonsuo, 2008)). b) summary of the effects of attention and conscious state on the magnitude of large-scale sensory brain responses. Specifically, results are depicted from two single-neuron recording studies that measured the effects of conscious perception (from (David A Leopold, Maier, Wilke, & Logothetis, 2005), modified) and attentional selection (from (Luck et al., 1997), modified) in the visual cortex of macaque monkeys. Note the similar pattern of enhanced responses co-varying with either phenomenon. Also note that that the magnitude of these enhanced responses varies considerably between cortical areas (i.e., increasing from early sensory cortex towards associative cortex). What is not shown are other changes at the single-neuron level that have been associated with the phenomena, affecting correlated activity, changes in neural noise, intra- and inter-areal coherence and changes in oscillatory activity, some of which cannot be observed when studying single neurons in isolation. Also note that these data sample neurons across entire cortical areas without regard to differences in cell type or neural circuitry (such as cortical layers) that they are embedded in (see (Maier et al., 2007; Sheinberg & Logothetis, 1997) for a sample of various types of single neuron correlate of consciousness). Thus, the neuronal concomitants of both consciousness and attention are likely more than a simple “boost” in neuronal activity, but rather constitute a complex arrangement of neural mechanisms with distinct regional characteristics.

Figure 4.

Promising new technological developments. a) Recent years witnessed a “quantum leap” in the count of simultaneously placed microelectrodes used in animal studies (from (Steinmetz et al., 2018), modified). b) This massive increase in the simultaneously measured activity of single neurons allows for more global (across-the-brain) measures at a sub-millisecond and sub-millimeter resolution. Given the spatio-temporal complexities associated with the neural concomitants of consciousness and attention (Figure 3), this unprecedented combination of macroscopic scale and microscopic resolution promises a host of new insights beyond current methodological constraints from (Stringer, Pachitariu, Steinmetz, Reddy, et al., 2019) and (Jun, Steinmetz, Siegle, Denman, Bauza, Barbarits, Lee, Anastassiou, Andrei, Aydin, Barbic, Blanche, Bonin, Couto, Dutta, Gratiy, Gutnisky, Hausser, Karsh, Ledochowitsch, Lopez, Mitelut, Musa, Okun, Pachitariu, Putzeys, Rich, Rossant, Sun, Svoboda, Carandini, Harris, Koch, OKeefe, et al., 2017), modified.

Figure 5.

Promising new theoretical directions. A possible research program for the search of an isomorphism between the structure of consciousness and the structure of information derived from the activity and structure of the neuronal activity. First, phenomenological structure needs to be precisely mapped out. As one possible mapping strategy, we can start with characterizing phenomenological relationships between various sounds (e.g., C major chords), as shown here, as well as conscious sensations in other modalities. Attention can be used here as a way to “perturb” phenomenology. These relationships can be quantified and mathematically expressed (e.g., in a multidimensional space through large-scale similarity rating experiments or in other ways using topology or category theory). Here, relationships among variously pitched tones are represented as a helix of musical “chroma” (Shepard, 1982), which captures the essential phenomenological feature of two tones that sound “similar but different”. At the neuronal side, connectivity and activity states of the neural networks are measured, and mathematically abstracted to propose physical structures (e.g., informational structure in IIT) that are supposed to give rise to the specific conscious experience. A search for an isomorphism starts with a unidirectional correspondence from one structure to the other (see main text). When both directions converge, we get closer to an isomorphic law that bridges between consciousness and brain activity. Attentional manipulation should preserve this isomorphism. That is, attention cannot change phenomenal structure without changing the information structure set up by neuronal activity, and vice versa. Brain icon is modified from James.mcd.nz, licensed under the Creative Commons Attribution-Share Alike 4.0 International, 3.0 Unported, 2.5 Generic, 2.0 Generic and 1.0 Generic license (https://commons.wikimedia.org/wiki/File:Brain_Surface_Gyri.SVG). The icon representing the information structure is modified from Takemori39, licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license (https://commons.wikimedia.org/wiki/File:Network_self-organization_stages.png).